Do I really Need The Anamorphic lens?

Re: Do I really Need The Anamorphic lens?

Well, yes and no, the issues are kind of related.

"Formatted to fit your screen" was a compromise surrounding "pan 'n' scan". Movies are 1.85:1, TV's are 1.33:1. They're simply not the same shape. So in the old days they'd pan 'n' scan the film, taking a subsection of the film and blowing it up to fill the full screen. If you watch a movie on TV and it doesn't have letterbox bars, you're basically missing at least 40% of the picture, and maybe up to 60% of it!

The anamorphic adapter changes the shape of the video. The DVX is 4:3, TV's are 4:3, but everyone seems to want 16:9. The anamorphic adapter bends light on the horizontal axis. It optically squeezes the horizontal image by 33%, so you record onto the full surface of the camera's CCD. You record a 16:9 signal by squeezing it into a 4:3 box, which then gets "unsqueezed" optically by a widescreen TV or digitally by a DVD player.

You don't get any "more" information, but you do get to use the full surface of the camera's CCD. If you shoot in "letterbox" mode or in "squeeze" mode, you only use about 75% of the surface of the chip. The anamorphic adapter lets you use 100%. But again, it's only for if you want widescreen footage, since that's what the adapter does: it makes 16:9 footage, just squeezed onto your 4:3 chip.
 
Hi Barry
Okay, I'm going to follow this thread a little further, if you don't mind.
Chris here, I'm a DP from way back, still own and run both spherical and anamorphic lenses in 35mm and also regular spherical in Super 16mm as well as my recent DVX100. I own and love your book by the way.
So... here's my question.

I totally get that the anamorphic adaptor compresses the image sideways into the full frame image, just the same as on film cameras.

How, if you're playing back on a 16:9 the pixels aren't the same shape and so the image comes out de-squeezed naturally. No problems there.

There is a $370 physical/optical viewfinder de-squeezer available for the camera, but I guess most people spend that money on a wide screen field monitor, because of focus issues and others. Okay.

Letterbox mode on the 100 and 100a basically is just the same as a hard matte on a camera gate - it chops around 25% of the image off top and bottom, simulating a 1:85 image. But now we are carrying only 75% of the full frame/pixel potential on to tape. Got that.

Now here's the part I don't really get.

If you have the full frame 4;3 information carrying the squeezed image from an anamorphic image, and you then desqueeze/downres to a letterbox on 4;3, doesn't that mean a higher resilution master creating a better downrezzed copy?

Or does the down-rez just eject the extra information, rather than using it in a compression algorhythm?

i.e. in theory a squeezed 4:3 image carrying 100% pixel power should reduce to a better image than an unsqueezed, letterboxed 4:3 image carrying only 75% pixel power, so to speak.

I know that in the end the pixels resolve down to 720 by whatever anyway, in either case. But is that extra 25% being put to some use? or is it just being discarded on the downres?

Isn't there some program that makes better use of the extra information if it is being discarded?

I know that optically my arguement makes all the sense in the world. A 16mm print from a 65mm original looks sharper, colors etc more accurately reproduced than a 16mm print from a 35mm or 16mm original, period, no matter what the granularity of the 16mm print stock.

That's because the edges of the molecules on the 16mm emulsion get a proportionately larger chance of getting the desired value transferred onto it from the larger original than from the smaller one.

(put another way, a postcard print off a 35mm camera neg looks less sharp than a postcard print from an 8x10 negative, no matter what stock that postcard print is made of).

So what about the electronic equivalent?

Eager minds want to know!!!
 
No.

Basically, you can't compare the optical transfer of a film print to the digital resampling of a higher-res image to a lower-res one.

I have this argument with HDV proponents all the time, who claim that starting with higher-res results in a better-looking DVD. It doesn't. I tested the living daylights out of it, under all conceivable circumstances, and the results are that it looks about the same, maybe a little softer.

Basically when you have a higher-res source that you downrez, you will end up blending pixels together, averaging them, and ending up with a slightly softer overall image than you would have gotten if you'd shot straight with the camera. In the letterbox example, you're talking about shooting a 720x372-pixel image, and leaving it as 720x372. Or, shooting that same image, but through an optical anamorphic adapter which stretches it, and then importing that into an NLE, digitally squashing/downrezzing it, and then compressing it again. No way does that amount of manipulation result in an overall better, sharper image. It's not much lower in quality, in fact it's pretty comparable, but on a strictly pixel-for-pixel basis, it is my contention that the letterbox image will still look better.

Pepole try to justify the oversampling/downrezzing argument by pointing to how much better film transferred to VHS looks than a VHS camera would look, and that's completely missing the entire point. The film isn't a digital file that's being resampled (with its attendant aliasing issues, blending, etc). It's a "reality" that's being photographed by a CCD (in the case of a Rank transfer) or scanned/digitized by a laser or whatever other method of telecine you choose to use. Versus an HD image that gets pixel-shrunk digitally. An entirely different process.

Still, it's basically easy enough to test -- shoot a letterbox scene without the anamorphic, then put the anamorphic on and adjust your framing so you shoot the same scene. Then stretch/crop the anamorphic scene to where it matches the letterbox and compare. I can't get to a test like that anytime soon, but maybe next week (unless someone wants to run the test for us now)...
 
Thanks Barry
So basically use anamorphic only when the optical arguement does apply? as in DI to 35mm print?
 
Barry
Thanks again for your input!

Also:

Are you implying in your arguement that the "blending together" of the pixels is strictly a fixed process by the mean (average) method, and that no other method to downres exists?

i.e. the sum-and-divide mean method is pretty crude, but is this ROM based, fixed?

Or is the blend/downres algorythm mutable?

If it isn't, don't you think there's room for improvement?

That's what I was getting at in my post.

I thought this stuff was digital, and therefore subject to the same kinds of manipulation that all digital (e.g. CG) imagery is.

For instance, a nearest-to adjacent (vertical as well as horizontal) blend (for lowest contrast, good for going to optical) or a most-different from adjacent (edge enhancement), and a whole lot of other algorythms I can think of to try.

Isn't there some software that does this already? (Besides Lightwave/Softimage/Maya, which is the way I'm thinking of going if I try this out?)

Naturally at this stage it would be frame by frame batch processing, probably Targa by Targa, if the image came straight in via firewire, uncomressed (digitally) but optically (anamorphically) compressed.

The output would, I agree, probably not look the same as the original image (CG composited footage rarely does) but surely it won't always necessarily be worse than shooting straight 4:3 letterbox and doing a 1:1 clone?

Or are you saying that the inherent resolution of a 4:3 is just so low anyway that the various compression methods I've mentioned all add up to a hill of beans, and end up lost in a wash of aliasing/blending issues, no matter what the downres algorythm?

Yes, your last sentence will tell the tale: "Then stretch/crop the anamorphic scene to where it matches the letterbox and compare".

I'm waiting for my anamorphic adaptor so I can do just that. Should be interesting.
 
You must log in or register to reply here.
Back
Top